Finite-sample analysis of M-estimators using self-concordance
This provides theoretical guarantees for statistical learning in misspecified models, but it is incremental as it builds on existing self-concordance theory with local assumptions.
The paper tackles the problem of finite-sample analysis for M-estimators by using self-concordance to determine the critical sample size needed for chi-square type bounds on excess risk, achieving bounds like O(d * d_eff) under minimal assumptions and improved to O(max{d_eff, d log d}) with stronger conditions.
The classical asymptotic theory for parametric $M$-estimators guarantees that, in the limit of infinite sample size, the excess risk has a chi-square type distribution, even in the misspecified case. We demonstrate how self-concordance of the loss allows to characterize the critical sample size sufficient to guarantee a chi-square type in-probability bound for the excess risk. Specifically, we consider two classes of losses: (i) self-concordant losses in the classical sense of Nesterov and Nemirovski, i.e., whose third derivative is uniformly bounded with the $3/2$ power of the second derivative; (ii) pseudo self-concordant losses, for which the power is removed. These classes contain losses corresponding to several generalized linear models, including the logistic loss and pseudo-Huber losses. Our basic result under minimal assumptions bounds the critical sample size by $O(d \cdot d_{\text{eff}}),$ where $d$ the parameter dimension and $d_{\text{eff}}$ the effective dimension that accounts for model misspecification. In contrast to the existing results, we only impose local assumptions that concern the population risk minimizer $θ_*$. Namely, we assume that the calibrated design, i.e., design scaled by the square root of the second derivative of the loss, is subgaussian at $θ_*$. Besides, for type-ii losses we require boundedness of a certain measure of curvature of the population risk at $θ_*$.Our improved result bounds the critical sample size from above as $O(\max\{d_{\text{eff}}, d \log d\})$ under slightly stronger assumptions. Namely, the local assumptions must hold in the neighborhood of $θ_*$ given by the Dikin ellipsoid of the population risk. Interestingly, we find that, for logistic regression with Gaussian design, there is no actual restriction of conditions: the subgaussian parameter and curvature measure remain near-constant over the Dikin ellipsoid. Finally, we extend some of these results to $\ell_1$-penalized estimators in high dimensions.